INFORMATION TO USERS

This reproduction was made from a copy of a document sent to us for microfilming. While the most advanced technology has been used to photograph and reproduce this document, the quality of the reproduction is heavily dependent upon the quality of the material submitted.

The following explanation of techniques is provided to help clarify markings or notations which may appear on this reproduction.

1. The sign or "target" for pages apparently lacking from the document photographed is "Missing Page(s)". If it was possible to obtain the missing page(s) or section, they are spliced into the film along with adjacent pages. This may have necessitated cutting through an image and duplicating adjacent pages to assure complete continuity.

2. When an image on the film is obliterated with a round black mark, it is an indication of either blurred copy because of movement during exposure, duplicate copy, or copyrighted materials that should not have been filmed. For blurred pages, a good image of the page can be found in the adjacent frame. If copyrighted materials were deleted, a target note will appear listing the pages in the adjacent frame.

3. When a map, drawing or chart, etc., is part of the material being photographed, a definite method of "sectioning" the material has been followed. It is customary to begin filming at the upper left hand corner of a large sheet and to continue from left to right in equal sections with small overlaps. If necessary, sectioning is continued again-beginning below the first row and continuing on until complete.

4. For illustrations that cannot be satisfactorily reproduced by xerographic means, photographic prints can be purchased at additional cost and inserted into your xerographic copy. These prints are available upon request from the Dissertations Customer Services Department.

5. Some pages in any document may have indistinct print. In all cases the best available copy has been filmed.

University MicrOfilms

International 300 N. Zeeb Road Ann Arbor, MI48106

8526324

Van Milligen, Fred Joseph

IN·SITU MONITORING OF THIN FILM GR0VviH USING A WIDE·BAND SCANNING MONOCHROMATOR

The University of Arizona

University Microfilms

International 300 N. Zeeb Road, Ann Arbor, M148106

PH.D. 1985

PLEASE NOTE:

In all cases this material has been filmed in the best possible way from the available copy. Problems encountered with this document have been identified here with a check mark_";_.

1. Glossy photographs or pages __

2. Colored illustrations, paper or print __

3. Photographs with dark background __

4. Illustrations are poor copy ",/

5. Pages with black marks, not original copy __

6. Print shows through as there is text on both sides of page __

7. Indistinct, broken or small print on several pages v 8. Print exceeds margin requirements __

9. Tightly bound copy with print lost in spine __

10. Computer printout pages with indistinct print __

11. Page(s) lacking when material received. and not available from school or author.

12. Page(s) seem to be missing in numbering only as text follows.

13. Two pages numbered . Text follows.

14. Curling and wrinkled pages __

15. ~her ____________________________ _

University Microfilms

International

IN-SITU MONITORING OF THIN FILM GROWTH

USING A WIDE-BAND SCANNING MONOCHROMATOR

by

Fred Joseph Van Milligen

A Dissertation Submitted to the Faculty of the

COMMITTEE ON OPTICAL SCIENCES (GRADUATE)

In Partial Fulfillment of the Requirements

For the Degree of

DOCTOR OF PHILOSOPHY

In the Graduate College

THE UNIVERSITY OF ARIZONA

1 985

THE UNIVERSITY OF ARIZONA GRADUATE COLLEGE

As members of the Final Examination Committee, we certify that we have read

the dissertation prepared by _____ F_r_e_d __ J_. __ V_a_n __ M_i_l_l_i~g_en ____________________ ___

entitled In-Situ Monitoring of Thin Film Growth using a \\I'ide-band --------------------~------------------------~------------------

Scanning Monochromator

and recommend that it be accepted as fulfilling the dissertation requirement

for the Degree of Doctor of Philosophy ----------------------~~-------------------------------

J-'.O.~ Date

!Jig! Jr:s-

Date I

Date

Date

Final approval and acceptance of this dissertation is contingent upon the candidate's submission of the final copy of the dissertation to the Graduate College.

I hereby certify that I have read this dissertation prepared under my direction and recommend that it be accepted as fulfilling the dissertation requirement.

Dissertation Director

STATEMENT BY AUTHOR

This dissertation has been submitted in partial fulfillment of requirements for an advanced degree at The University of Arizona and is deposited in the University Library to be made available to borrowers under rules of the Library.

Brief quotations from this dissertation are allowable without special permission, provided that accurate acknowledgment of source is made. Requests for permission for extended quotation from or reproduction of this manuscript in whole or in part may be granted by the head of the major department or the Dean of the Graduate College when in his or her judgment the proposed use of the material is in the interests of scholarship. In all other instances, however, permission must be obtained from the author.

SIGNED: /~ JryL1--&-----

ACKNOWLEDGMENTS

As is the case in most modern research much of this project

could not have been done without the help of others. I would like to

thank the people who both helped me in this work and those who were

influential in my arriving to this point in my career.

It would be impossible for me to overstate my gratitude to H.

Angus Macleod, my dissertation advisor, for all the help he has given

me over the last five years. He has built a research group at the

University of Arizona which is outstanding in its achievements, but just

as important, a pleasure to be part of. I can only hope that his

gentlemanly demeanor, which permeates our group, will follow all of us

as much as the knowledge he has passed along.

The thin film laboratory at the Optical Sciences Center has

been blessed- with two individuals who are very accomplished at keeping

it running, even with the constant turnover in graduate students. I am

indebted to Ross Potoff for all the help he has given me in the

construction of this system as well as imparting knowledge on the

mechanics of the thin film industry; he is, perhaps, the perfect

compliment to Angus in our education. I would also like to thank Mike

Jacobson for his ability to keep things organized, while we all did our

best to combat his efforts, and for his helpful suggestions over the

years.

There have been people who' have helped immensely in the

construction and application of the scanning monochromator system. I'm

iii

iv

deeply grateful to: Bertrand Bovard whose work during his one year

post-doc should really entitle him to be co-author of this

dissertation, Jim Mueller who was able to develop the electronics for

the system well enough that I've been unable to ruin them, Professor

Richard Shoemaker for all of his help in the early programming of the

system, and Emile Pelletier and Francois Florey for the inspiration of

building the system, as well as for many helpful discussions.

I would also like to thank everyone that I've worked with in

the lab; I feel I've learned something from each of them. I'd like to

single out a few with whom I've worked especially closely. They are:

Steve Saxe, who proceeds me in his defense by a week, Mike Messerly

who has both helped me in the lab as well as put up with me as a

roommate while this was being written and Steve Browning for training

me when I first arrived here before he completed his dissertation

requirements. I am also grateful to Professors Ursula Gibson and

Bernhard Seraphin for their help in the completion of this dissertation.

I also thank Marcy Osgood for helping with some of the figures and Lisa

DuBois and Anna McKew for preparation of the final format, as well as

for their friendship.

On a more personal note I'd like to thank Paul Atcheson, Terry

Ferguson and Sean Keck for sharing beers and friendship and in the case

of Paul, his occasional calling of a good game behind the plate while I

pitched. Finally I'd like to thank my family for al~ays supporting me

in my decisions over the years, and. Ann Fasanella for bringing me such

happiness this last year.

TABLE OF CONTENTS

Page

LIST OF ILLUSTRATION& vii

LIST OF TABLES. ....................................... ix ABSTRACT. • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • •• x

1. INTRODUCTION. • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • •• 1

2.

3.

4.

Why are Thin Films Important? ••••••••••• Optical Properties of Thin Films ••••••••• Use of Optical Thin Films in Fil ter Design. Parameters in Thin Film Deposition ••••••••

THICKNESS MONITORING OF THIN FILMS DURING DEPOSITION ••

Physical Monitoring Techniques •••••••••••• Optical Monitoring •••••••••••••••••••••• Advantages of Physical or Optical .Monitoring

THE SCANNING MONOCHROMATOR SYSTEM •

Sys tem Design. ••••• The Light Source ••• The Optical System.. The Detector •••••••

. . . . . . . . . . . . . . . . . . . . . . .

Computer Handling of Data ••••• Overall System Performance. ••• Applications of the System •••••

DERIVATION OF OPTICAL CONSTANTS FROM SMS DATA ••

The Envelope Method. •••••••••••••• Determination of N(d). •••••••• Determination of Thickness d. ••• Determination of k. ••••.••••••• Smoothing of Transmission dat~ Drawing of Envelope ••••••••••••••• Computer Simulation of Thin Film Deposition

and Application of Analysis Technique. •••• Application of Technique to Titania Films •• Limitations of the Technique ••••••••••••••••

v

2 4 8

12

15

16 22 27

31

31 32 35 37 39 41 42

51

52 54 55 56 56 57

61 62 66

5.

vi

TABLE OF CONTENTS--Continued

Page

SUMMARY •••••••••••••••••••••••••••

Extension into the Ultraviolet •• Ultraviolet Studies •• Conel usion •••••••••••••••• . .

APPENDIX A: COMPUTER HANDLING OF SYSTEM

APPENDIX B: ELECTRONICS ••••••••••••••

. .............. . . .. . . . . . .. . .

..................

.................. REFERENCES • . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

68

70 76 76

78

98

116

LIST OF ILLUSTRATIONS

Figure Page

1.1 Electron Micrograph of Thin Film •••••••••••••••••••••••••• 5

1.2 Computer Simulation of Thin Film Growth •••••••••••••••••••• 6

1.3 Comparison of Packing Density Definitions ••••••••• 9

1.4 Simple Model of a Thin Film •••••••••••• ................. 11 2.1 Thickness measurement of a metal film by

Monitoring its Resis tance • • • • • • • • • • • • • • • • • • • • • • • • • • • • •• 17

2.2 Microbalance Measurement of Thin Film Mass •••••••••••••••• 19

2.3 AT Cut Quartz Crystal ••••••••••••••••••••••• 21

2.4 Typical Quartz Crystal Monitor Layout •••••••••• 23

2.5 Optical Monitoring System •• ........................ 26 2.6 Reflectance Measurement of Film During its Deposition.. • • • • •• 28

2.7

3.1

3.2

3.3

3.4

3.5

3.6

3.7

3.8

Optical Monitoring Signal of Multilayer During Deposition •••••

Scanning Monochromator Flow Diagram ••••••••• . . . . . . . . . . . . . a) Spectral Profile of Tungsten Halogen Lamp b) Spectral Profile of Xenon Arc Lam p ••••••••••• . . . . . . . . . Appearance of Scanning Monochromator System ••••••

Top View of Scanning Monochromator System •••••••• . . . . . . . . . Flow Chart of Computer Data Handling Program..

Cary 14 Transmission Trace of Didymium Glass ••

Waveleng,th Calibration Plot •••••••••••••••••••

Example of Water Adsorption in Ti02, Si02 Fabry-Perot Filter ••••••••••••••••••• .................

29

33

34

36

38

40

43

44

46

3.9 Example of Reverse Monitoring Trace •••••••••••••••••••••• 49

vii

LIST OF ILLUSTRATIONS--Continued

Figure

3.9 Example of Reverse Monitoring Trace

4.1 Kalman Filter Used to Smooth Data ••

4.2 Plot of Noisy Signal ••

4.3 Fil tered Signal •.........•••••••••.•.••..••.•••

4.4 Profile of Refractive Index and Extinction Coefficient for a Stable Titania Layer. (Upper curve represents N,

Page

49

58

59

60

lower curve K)....... . • • .. • • • • • • • • • • • • . • • • • . . . • • . • • • .. 63

4.5 Dispersion of Innermost and Outermost Refractive Index for a Stable Layer of Titania Film.. • • • • • • • • • • • • • • • • • • • •• 64

4.6 Example of Result Given by Method When Applied to an Uns table Layer ••••••••••••••••••••••••••••.••• 65

5.1 Overall View of System Including UV System. •••••••••••••••• 71

5.2 Top View of UV System ••••••••••••••••••••••••••• 72

5.3 a) Spectral Response of Redcon Array, window removed. b) Spectral Response of Redcon Array, quartz window ••• 74

viii

LIST OF TABLES

Table Page

2.1 Film color as a function of optical thickness for ZnS and Na3A1F6..................................... 24

3.1 System Performance •••••••••••••••••••••••••••••••••• o. 41

ix

ABSTRACT

To augment the monitoring capabilities of a Balzers 760 coating

chamber, we replaced the simple, single wavelength optical monitor with

a wide-band scanning monochromator system which records transmission

data over the visible region of the spectrum. The system is controlled

by an IBM-PC. The same computer is also interfaced to a quartz crystal

monitoring system which was purchased with the Balzers chamber. The

scanning monochromator system required a new brighter light source to

deliver sufficient signal to the detector array through the more

complex, dispersive optical train. Above the chamber the filter and

the photomultiplier pair were removed, and replaced by a flat mirror

which diverts the beam horizontally into the scanning monochromator

system. The beam passes first through a telescope-slit configuration

onto a Jobin-Yvon holographic grating, built to disperse the 400-800 nm

band of which we use approximately 360 nm. This reflective grating

images the spectrum of the slit onto a Fairchild CCD array, which

consists of 1728 elements. These elements are then averaged into 173

data points and recorded by the IBM-PC. The 173 data points allows us

a wavelength resolution of about 2 nm. The IBM incorporates a Tecmar

AID board in accepting data from both the quartz crystal monitor and

the scanning monochromator system. Although the system is capable of

recording data at a faster rate, it is generally stored once every

three seconds. This is adequate since at normal depOSition rates this

gives us information every 10 - 20 Angstroms of deposited material.

x

xi

The system has been used in several applications which will be

discussed in this dissertation.' They include in situ measurements of

water adsorption into a film, derivation of optical constant profiles

during the film deposition, both of which may lead us to a better

understanding of the growth of a thin film. The monochromator has

also been used to analize the components of a multilayer coating by

monitoring the film's transmission spectra while it was sputter-etched

off. The extension of the system into the ultraviolet region of the

spectrum and some future applications are also considered.

CHAPTER 1

INTRODUCTION

The current state of research in optical thin films has

progressed.to the level where the potential of the traditional empirical

approaches have been all but exhausted. Further progress in the studies

of thin films demands the use of more fundamental methods of both

analysis and evaluation. Essential to this drive towards a more basic

understanding is the existence of more detailed and meaningful

measurements. Of particular importance to the future understanding of

the growth of thin films is the capability of recording the evolution of

the optical properties during deposition. This dissertation will describe

in detail the conception and realization of an instrument to fulfill this

task. Although the motivation for developing this instrument was to

allow basic research into the structure of thin films there are other

applications of this system considered also.

Optical coatings require very accurate control of their layer

thicknesses and optical properties during deposition. This process

control is referred to as monitoring. There are still great problems

and barriers to progress in this area. The instrument which will be

described has great potential in the monitoring area, as well as in the

fundamental studies of optical properties. Therefore, information

concerning both uses of the instrument will be discussed in this

dissertation •. Before describing this instrument in detail it is

1

2

appropriate to first discuss some of the properties of thin films and to

review the current state of monitoring processes. The first two

chapters of the dissertation will be devoted to these topics.

Why Are Thin l"ilms Important?

The most important aspect of optical thin films is their

performance. They greatly improve the characteristics of almost any

optical system in a manner that is unachievable in any other way. Their

first important application was as a simple antireflection coating for

binocular optics. It was found that the binoculars were able to work at

much lower light levels after a single thin film layer was applied.

Another important feature of optical thin films is that they can be

deposited directly onto most components which need them without

appreciably changing the size or shape of the component. This saves

space within the system and also alleviates the worries of redesigning

the mounting systems of the assembly. Since the coatings operate by

interference effects, they are very thin and therefore can be deposited

fairly simply in transparent form and with layer boundaries that are

both smooth enough and close enough to parallel that interference

effects can be seen.

This use of thin films as optical fil tel'S has been known to

most, though they may not have realized it, since their childhood. The

ability of films to display a spectral profile has been noted by most

children while blowing soap bubbles. The colors of the bubbles are

caused by inte~ference effects as a consequence of the varying thickness

of the soap film the bubble is made of, and, indeed, the colors of soap

3

films were discussed in detail by Sir Isaac Newton in his Opticks. Of

course, if this was the extent of man's ability to use optical thin films

the literature pertaining to the subject would have ended there.

Fortunately, the situation is quite different. It is possible, through

use of a multilayered stack of films, to design filters of very precise

spectral profiles. In fact, assuming an infinite selection of materials,

which of course in practice is not the case, it has been shown that a

multilayer stack can be created to give any desired spectral profile

(Dobrowolski 1978).

A further major advantage of thin films, which will also be

presented, is the possibility of tailoring a film in ways difficult or

impossible with bulk materials. Better properties can be obtained by

creating films which are mixtures of materials. An example of this was

described by Pellicori in 1984. He was able to reduce the stress in a

CeF: film by coevaporating it with other fluorides. Another way of

tailoring a thin film's properties which has received much attention

recently is by ion bombardment during evaporation. Botten et ale (1984)

have shown the ability of attaining a wide variety of refractive indices

of TiO: films by careful control of the ion bombardment process. This

ability to manipulate the properties of a thin film often requires the

capability of measuring the characteristic of the film while it is

growing.

There is, however, another side of the picture. The differences

between film and bulk properties are not all favorable. Many aspects of

the behaviour of thin films cause great problems both during and after

deposition. Further advances in the field of optical coating demand

4

improvement and can come only with better understanding. The instrument

described here is intended to contribute to this search. The next

section describes briefly our current knowledge of the real optical

properties of thin films.

Optical Properties of Thin Films

The properties of a thin film material may differ greatly from

those of the bulk material. This variance from the bulk properties are

caused primarily by the different microstructure of a thin film. Films

deposited by physical vapor deposition tend to have a columnar

structure, unlike that of the bulk material. Figure 1.1 and Figure 1.2

are given to further represent what is meant by the columnar structure

of a thin film. Figure 1.1 is a scanning electron micrograph of a thin

film and Figure 1.2 is a computer simulation of a thin-film growth

process (11ao 1985). One other cause of the variance of the properties

of thin films to those of bulk is the strong surface effects in a thin

film. Unlike bulk much of the material in a thin film layer is located

near one of its two surfaces. These differences allow thin films to

have unique properties, not all of which are favorable.

The most important parameter in the design of optical thin film

filters is the optical thickness of the layer. It is this parameter that

defines the phase of the wavefront entering or leaving the layer; thus

controlling the interference effects of the layer. This thickness is

defined to be the physical thickness of the layer multiplied by the index

of refraction of the material. The index of refraction, n, is a property

of the particular layer deposited. It is defined as:

j __ _ ·----L

: .. 1 ".

-· .:/ , ~ j :,•)"

Figure 1.1. Electron Micrograph of Thin Film

Q) u m 't :J

(f)

ZnS

ZnS

6

Figure 1.2. Computer Simulation of Thin Film Growth

n=c v

7

where c is the speed of light in vacuo and v is the speed of light in the

material. If the material is a perfect dielectric, this would be all

that need be considered in a first order design. Unfortunately, in the

re'al world materials absorb some of the light passing through them. A

measure of the absorbing property of a film is its extinction

coefficient, k. The two parameters, (n,k), are commonly referred to as

the optical constants of the material. Values of optical constants

should be treated very carefully. These optical constants can vary in

respect to both wavelength and thickness and are very dependant upon

deposition parameters. It is also important to note that the values of

nand k can vary greatly in a thin film from those that have been

established for similiar bulk material. This variance from bulk

properties can have several causes including a possible change in

structure (for example Gary Carver, 1979, has shown that it is possible

to deposit molybdemum films that are fcc instead of bcc) of the

material and the simple fact that the films will almost surely contain

some internal voids originating in the growth process.

A measure of the voids contained in the film is given by the

packing density of the film, p.

p volume of solid material/(volume of solid material

plus volume of internal voids).

Packing density is related to the optical constants of the film. In a

given dielectric film, the lower the packing density the lower is the

8

refractive index. The relationship between these two quantities is not a

simple one. The particular form of the microstructure is involved as

well as the packing density. A simple relationship that is a linear

interpolation between extremes is:

n=nbulkP +(l-p)nvoid

This is used often because of its simplicity, although it can be

considerably in error in the case of high-index films. There are great

difficulties in accurately measuring p and an alternative method which is

frequently used is to invert the previous equation to give:

n-nvoid p= nbul k-nvoid

The packing density calculated in this fashion is not the true packing

density but can be used in comparisons between films produced under

different process conditions.

The relationship between p and n has been studied in greater

detail by Harris et al. (1984) (1979). In this work she was able to show

for values of p an expression for n given by Bragg an~ fippard (1953) is

more correct:

n = L «(1-p)n"·void+(1+p)n2 void n2S)] 1/2 l (1+p)nz·void+(l-p)nZ s A comparison of the two definitions of n is presented in Figure 1.3.

Use of Optical Thin Films in Filter Design

The interference effect caused by thin films has been modeled

sufficiently well that they may be utilized in the production of filters

of well defined spectral performance. By applying Maxwell's equations

and boundary conditions on the continuity of fields across the layer

9

1·0

0·8 ns =2·35 /

:0 0·6 / ~ / a :> / ~ 0·4 / ~ - / 0.0·2 /

a 0 0·2 0·4 0·6 0·8 1· a

p (actu al)

4·0

b

.2·0 c:

o 0·2 0,4 0·6 0·8 1·0 p (actual)

Figure 1.3. Comparison of Packing Density Definitions

10

boundaries, it can be shown that a parallel sided homogeneous layer(see

Figure 1.4) may be modeled mathematically by the following matrix,

commonly referred to as the characteristic matrix (Macleod - Optical

Thin Film Filters):

COSO l

21f(n-ik)coseld n-ik where 15 1= A and n=(n-ik)cose l for TE waves or cose l

for TM

waves. From this characteristic matrix the performance of any

multilayer filter may be determined by the following matrix equation:

where Mi is the lcJ = : [J [J

characteristic matrix for layer i.

the multilayer can then be determined as

with y=.f B·

n -y R=(_O_)2

no+Y

The reflectance of

Inspection of the characteristic matrix revealR the importance

of layers that have optical thicknesses that are multiples of {. These

layers, which are commonly called quarterwaves or half waves, tend to

be the building blocks of most thin film designs. It is at these

thicknesses that the reflectance or transmittance of a single layer upon

a substrate has an extremum. It can clearly be seen from the

characteristic matrix that a halfwave layer will not have any effect at

the design wavelength, since at this thickness the characteristic matrix

11

"air

Figure 1.4. Simple Model of a Thin Film

12

reduces to the identity matrix. In the case of quarterwave thicknesses,

y= nsubstrateCos/)l + inlsin/)l + i( nsubstratesin/)l) COS/)l n1

which reduces to the extreme value for Y,

n12

Y=-----nsubstrate

at any odd multiple of a quarterwave. This approach permits design of

advanced stacks of demanding performance.

Parameters in Thin Film Deposition

There are a large number of process parameters involved in the

deposition of thin films. For physical vapor deposition these include

substrate temperature, vacuum conditions, starting material, how often

the starting material has been heated, rate of deposition and substrate

preparation. Each of these can influence the microstructure or the

composition of the film, which in turn creates the optical properties.

These parameters are capable of effects from simple changes in the

packing density of the film to major modifications in the entire

structure of the film (an example of this is found in Ti02 which can be

deposited in the form of Anastase or Rutile, depending upon deposition

techniques - Pulker, Paesold and Ritter, 1976).

Because of the large number of parameters to be controlled in a

deposition it is economically important to find a fast way of determining

the best process parameters. This requires a monitoring system which

will allow one to determine those properties of the film deemed most

important for the application. For optical thin films the most

important properties to control are the films thickness and optical

13

constants. Many schemes have been devised for monitoring thin film

deposition, (Some of these will be covered in more detail in Chapter 2),

but almost invariably they have been designed only to monitor the

thickness of the film, leaving the determination of the optical constants

(n,k) for analytical techniques after deposition, and generally after

removal of the sample from the coating chamber, implying the

measurement of these constants only after they have been exposed to

atmospheric conditions and the determination of an average value of n

and k over the entire film thickness. This empirical approach to the

understanding of thin film processes involving the study of the effects

of process changes on macroscopic properties only, is what we intend to

replace. We seek a better understanding of what is actually happening

to the microstructure of the film when one of the process parameters is

changed.

For this, it is imperative that measurements be performed in

situ rather than post deposition, as previously. The optical properties

so measured will be used as a probe of thin film microstructure. The

evolution of the optical constants as the film grows is essential data

for understanding the evolution of the microstructure of the film. The

scanning monochromator system's ability to take large quantities of data

during deposition makes these measurements possible. It is also very

important to be able to measure the voids in a film. This is usually

done by measuring the adsorption of water by the film after it has been

exposed to atmosphere. This can be done by the scanning monochromator

system without removing the sample from the chamber by watching the

shift in the spectral profile. This is only possible because of the

14

systems wide-band information. Although the strengths of such a system

have long been known, it has not been technologically feasible until

recently. Similar systems have been created with displays on a cathode

ray tube, but these systems were incapable to do any form of data

manipulations and therefore were very limited in their usefulness. With

the recent advances of minicomputers, it has become possible to acquire

data, while still allowing some calculation capabilities, fast enough to

allow systems such as the scanning monochromator to approach their full

potential.

CHAPTER 2

THICKNESS MONITORING OF THIN FILMS DURING DEPOSITION

The deposition of individual thin films has become, with the

advent of better vacuum systems, a relatively simple task. This

simplicity does not extend to the theory underlying the coating design

or the monitoring of the thin film deposition with sufficient accuracy

to create the desired film. The most important prnperty for control in

an optical thin film coating is the optical thickness. This determines

the phase lag that occurs in the wavefront which then leads to the

desired interference effects. ThE' accuracies required in the layer

optical thicknesses vary with the type of filter under construction,

and often upnn the particular layer in the coating the error occurs in,

but tolerances may be as low as two percent. There are several

different ways to measure the optical thickness of a film during

deposition. They fall into two broad categories, which we will call

physical .monitoring and optical monitoring.

These divisions are based on the parameters that each method

actually measures. Physical monitoring consists of the various

techniques which measure the mass or thickness of the material and

then rely upon a knowledge of the optical constants of the film to

derive the optical thickness. Optical monitoring entails those

techniques which directly measure the optical thickness of the layer.

This chapter will be divided into three parts, two describing each of

15

16

the two types of monitoring and a third section concerning the

advantages and disadvantages of the two methods.

Physical Monitoring Techniques

Physical monitoring techniques may rely upon several different

properties of the film, such as the electrical properties of the

material, momentum of the evaporant or the weight of the deposited

film. Each of these may be more or less useful as the basis for

monitoring depending upon the coating being deposited. In the case of

optical thin films the most widely used technique is based on the

measurement of the mass of the material deposited on a piezoelectric

quartz crystal which forms part of an oscillating circuit. Before

considering quartz cystal monitoring in detail on this method, I will

briefly describe the others.

There are several different types of monitoring which are based

upon the electrical properties of the material. Since these properties

may vary greatly, depending upon the material, there are quite

different approaches to this class of monitoring. For metal films, it

is possible to measure the electrical resistance during deposition.

The metal film can be placed in one arm o~ a wheatstone bridge,

calibrated such that the correct thickness of the film will balance the

bridge (Bennett and Flanagan, 1960, see figure 2.1). In contrast to

this, in a dielectric film it is possible ·to measure a change in the

capacitance across parallel plates as the dielectric is deposited

between them (Keister and Scapple, 1962). Both of these techniques

require a precise knowledge of the film's electrical properties in

VAPOR SOURCE

REFERENCE RESISTOR

SHU1"TER SOLENOID

RECORDER CONTRO\..LER

A"P\.IFIER

BRIDGE T VOLTAGE I

1'7

Figure 2.1. Thickness measurement of a metal film by Monitoring its

Resistance

18

order to relate the measurement to the physical thickness, which is

still once removed from knowledge of the optical thickness of the

layer.

Another way of monitoring the thickness of the film is to

measure the deposition rate of the evaporant as well as timing the

length of the run. Two possible techniques are to measure either the

physical momentum of the evaporant or the ionization current in the

chamber caused by the presence of the evaporant. With this information

and a very strong knowledge of the system geometry, background

atmosphere and sticking coefficient of the evaporant to the substrate

it is possible to determine, or at least approximate, the thickness of

the film on the substrate. These methods are rarely used in optical

coating deposition.

More useful in the production of optical coatings are methods

which directly measure the mass of the material being deposited by

means of a microbalance. This class includes the quartz crystal

monitor technique. There are many different types of microbalances

which have been incorporated into thin film vacuum chambers. Figure

2.2 shows an example of one such system (Mayer, Schroen and Steunkel

1960). Here the mass deposited on a vane is measured by a restoring

force applied by a solenoid and magnet.

copper cylinder damp oscillations in

Eddy currents induced in a

the system. Although

micro balances are able to measure very small mass with very high

accuracy, for thin film deposition their precision unfortunately suffers

greatly from mechanical vibrations, as well as electrostatic effects

(Behrnd t 1956).

SOI.ENOID~ Cu-CYI.INDER -~!:L1'lI

SUSPENSION " FIBER"

, SPRING

CAI.18RATION PAN

VANE

Figure 2.2. Microbalance Measurement of Thin Film Mass

19

20

These problems are largely overcome in the quartz crystal

microbalance found today in most modern coating chambers. In fact the

coating chamber used in the research to be described is equipped with a

microprocessor which bases its decisions on the data generated by the

quartz crystal. The quartz crystal microbalance uses the thickness

shear mode of a piezoelectric quartz crystal. The crystal is in the

form of an AT cut crystal (Fig 2.3) since this geometry has the lowest

temperature dependance near room temperature. Because the natural

resonant frequency of the crystal is mass dependant, it is possible to

determine the mass deposited on the crystal by monitoring its natural

frequency. This is done by using the crystal, complete with electrodes

as the controlling element of an electronic oscillator. The resonant

frequency of the crystal is given by

where N = 1.67 x 106 hz mm and d is the crystal thickness.

This expression for the resonant frequency holds as long as the

deposited mass is small in comparison to the crystal mass. The change

in frequency with deposited mass is given by:

where Cf is a constant of the crystal, K is approximately 1 and Pq is

the crystal density.

It is important to note that ~f is a function of fo~ therefore,

more sensitivity is possible for crystals with higher natural resonance

frequencies. Unfortunately the requirement that the deposited film be

thin in comparison to the crystal implies crystals with higher

21

/ i \,

I

Figure 2.3. AT Cut Quartz Crystal

22

frequencies, which must be thinner will not last as long as lower

frequency less sensitive ones.

In practice, quartz crystals are often employed in pairs. One

shielded from the evaporant, is used to provide a reference frequency

while the other is deposited on. In this way it 1s possible to measure

the change in frequency as a beat frequency between the two vibrating

crystals, which can be measured much more accurately. The crystal

head is normally water cooled to retain the crystals in the zone of

low temperature sensitivity. The entire system is very rugged and

quite well suited to vacuum applications. (An example of a typical

quartz crystal monitoring system is given in Figure 2.4).

Optical Monitoring

Although it is not a universal solution to all thin film

monitoring problems, optical monitoring is an extremely valuable tool

in the control of optical thin film deposition. It allows the direct

measurement of the most important parameter, optical thickness. This

section will discuss three different methods of optical monitoring,

visual monitoring of a film's color, polarimetric methods and

reflection or transmission measurements.

The first, and perhaps simplest monitoring technique is to

watch the color of the film change as it grows on the substrate. This

procedure, though at first sight appearing inaccurate, performs

surprisingly well for some single layer coatings, especially the

standard MgF2 antireflection coating for glass. Table 2.1 gives an

example of monitoring colors for both a low and high index coating

(Banning 1947).

COUNTER

I I

MIXER ~

I : y~ MIXER ~

I PULSE

SHAPING

RECO~OER OR METER

SHUTTER RELAY

Figure 2.4. Typical Quartz Crystal Monitor Layout

23

Table 2.1 Film color as a function of optical thickness for ZnS and Na3AIF,

24

Color Change Optical Thickness

(A=SSOnm)

ZnS

(n=2.35) (n=1.3S)

Bluish White Yellow

White Magenta A 4

Yellow Blue

Magenta White A 2:

Blue Yellow

Greenish White Magenta 3A T

Yellow Blue

Magenta Greenish White

Blue Yellow

Green Magenta SA T

25

Although the perception of color in a thin film thickness determination

is subjective, it can be made quite sensitive in some cases by proper

choice of monitoring substrate. An example of this was reported quite

recently by Sandstrom, Stenberg and Nygren (1985). They report a

technique of enhancing the interference color caused by very thin

organic films by growing them on a Si substrate previously treated with

a very thin Si02 layer.

The thickness of a thin film can also be determined by

polarimetric methods. Drude first demonstrated that one can determine

the innex and thickness of the film by measuring the change in

polarization state of the light reflected by a thin film. This

technique has proven to be one of the most accurate ways of

determining these quantities. Unfortunately, the complexity of the

system and of the mathematical analysis necessary to e?ttract this

imformation from the data, has made its application rlifficult. Because

of their complexity, in-situ elipsometric monitoring systems have gained

very little support. Some work has been reported in Nature (1951,

Hermanson) and more recently an outline of the requirements for such

a system was reported in Surface Science (1973, Jasperson, Burge and

O'Handley). The reader is referred to these articles for more details

on the subject.

The most commonly used method of optical monitoring, due to

its simplicity and its wide applicability, is the photometric

measurement of reflection and/or transmission data during deposition.

A typical arrangement for this measurement is shown in Fig 2.5. A

narrow wavelength band of· light is selected from a modulated white

I C

/ !

Two-beam principle

;:

i; ii

--.I.. ---rr-

;1 I;

~~ GC~~B I . I

I I I

. A i I

Mono-beam principle

Measuring arrangement of the oPtical film thickness monitor BALZERS GSM 210. A Modulated light source E Test glass changer B Receiver Refl. . F Light beam deflection C Receiver Transm. G Cetlector C Indicating instruments H Intermediate-piece

Figure 2.5. Optical Monitoring System

26

27

light source by a narrow band filter. This light is then directed onto

the sample to be monitored which transmits or reflects it onto a light

detector. The optical thickness of the material can be determined from

the variation of the signal from the detector. This technique is well

suited to the deposition of dielectric materials since the signal should

be sinusoidal with each turning point corresponding to ~ucessive

quarterwaves of optical thickness at the monitoring wavelength. An

example of such a signal is given in Fig 2.6. This method works very

well for any coating design which employs alternating quarterwaves of

materials. An example of a monitoring signal for a coating consisting

of alternating ZnS, MgF2 is given in Fig 2.7. The process is less

advantageous for the monitoring of absorbing films since their optical

characteristics stabilize after a certain limiting thickness. A

thickness greater than this limit cannot be monitored accurately in

this manner.

Advantages of Physical or Optical Monitori,ng

The monitoring technique should be chosen carefully to suit the

film system under study. Any versa til e coating plant should be

equipped with both a quartz crystal monitor and some kind of optical

monitor, thus allowing the operator to enjoy the advantages of both

systems. After careful calibration, the quartz crystal monitor can be

used with great precision to measure the thickness of the layer being

deposited. Unfortunately, since the monitor receives no optical

information, for optical coatings it can allow errors to accumulate

without possibility of correction during the process. The use of a

conventional optical monitoring system has the, great advantage of

0,2

0,1 5 t--+-++-Ir--+++--";f---l-l--\-~

O,051+---+--~~---1I-----'_+I_-_I_-_l.I

o ;./2 'Ji-l. ). Optical thickness

n.: 2

n. = 1.4

Figure 2.6. Reflectance Measurement of Film During its Deposition

28

100~------------------------------~ }. =550 nm

9O~\--------------------~~~=7-t--~

t~~~-+~----~~~.-r-~~--+-~ .... II

g 70~-T~--~--------t-t---t---+---~--~-: ~ I E 60 ~-+---...I-.:rl+-t-----t--- 0 = l!. ~ ! 50 ~Oj(IIr;I':.f~.F' i ~40 : --1 II . I : (.) I n •. I. = il4. Glass = 301---++~-+~-4--~~~t---'-------'---1---1 n I I ! I II ,~ • ~ 20 \ a:: ,

\

"I "'. '" ~I ~I (/)1 ~i en ~: ~I o·

~ c. ~I NI ~~ N,

0 1 2 3 4 5 6 10 Number of the films

29

Figure 2.7. Optical Monitoring Signal of Multilayer During Deposition

30

measuring the desired optical performance directly. This gives a very

stable performance around the monitoring wavelength (Macleod 1972)

because of a natural error compensation process in quarterwave

monitoring, i.e., if one layer is slightly too thick the next layer will

reach its turning point at a thickness slightly lesser than it would

have, thus compensating for the previous error. This only holds near

the monitoring wavelength, so broadband coatings may not be corrected

adequately. The use of optical single wavelength monitoring also

presents difficulties for any thickness other than multiples of a

quarterwaves. In such cases a quartz crystal monitoring system may be

better. The inadequacies of both types of monitoring techniques have

motivated us to develop a system which incorporates both styles of

monitoring with an added enhancement of wide-band optical monitoring.

This enormously increases our ability to analyze the coatings we

produce.

CHAPTER 3

This chapter is divided into two major sections. The first

covers the design of the scanning monochromator system, while the second

discusses some applications either in our laboratories or reported on

elsewhere in thin-film literature. One such application of the system,

the determination of the optical constants (n,k) of a material is

mentioned only briefly in this chapter for completeness and then covered

in detail in Chapter 4.

System Design

Our scanning monochromator system (SMS) was designed to augment

the capabilities of a Balzers 760 box coater, which was originally

purchased with an automated process controller (Balzers Model KB 101)

based on data from a quartz crystal monitor. The system was also

equipped with a single wavelength optical monitor (Balzers Model GSM

210). The quartz crystal system was left intact and incorporated into

into the SMS. The single wavelength monitoring system was removed to

allow a pathway for the light of the wide-band transmission

spectrometer portion of the SMS. In the removal of this apparatus it

was found that the windows of the chamber were slightly misaligned,

causing significant vignetting. This was corrected by displacing the

incoming beam of light with two mirrors. These two mirrors are checked

periodically for both alignment and reflectance, but since they are well

31

32

shielded below the floor of the coating chamber, their reflectance has

been found to be constant throughout any coating run.

Figure 3.1 is a descriptive flow diagram of the scanning

monochromator system. Our addition, shown in the upper half of the

figure can be divided into four major components: the light source, the

optical system, the detector and the computer.

these parts will be discussed separately.

The Light Source

For clarity, each of

To ensure an adequate signal reaching the detector array, the

original Balzers light source had to be replaced. The new light source

had to meet the following criteria: enough light to be able to saturate

the detectors and sufficient stability to allow us to rely on a single

100 % reference line for an entire coating run. Our initial selection

was a 500 watt tungsten-halogen lamp. This source was found to have

some problems, but did work well enough to produce all of the

experimental information presented in this dissertation. Unfortunately,

the spectral profile (Figure 3.2a) varies rapidly with wavelength, and

does not extend at all into the ultraviolet region of the spectrum. In

fact, it proved impossible to arrange an appreciable signal between 400

and 440 nm without saturating the detectors at the red end of the

visible spectrum. In solving this problem we considered two possible

approaches. The first was to add a thin film filter in front of the

detectors to pass the blue end of the visible while attenuating the red.

Though this appeared feasible, we decided instead to change the light

source entirely.

HOLOGRAPHIC GRAT1NCI

8

~

> i-(I)

z L.I.J ~ z L.I.J > i-

35

The new light source in the system is a 1000 watt xenon-arc

lamp, which has a much flatter spectral performance and output in the

ultraviolet. (This spectral profile is shown in Figure 3.2b.) This light

source will allow us to extend the system into the ul traviolet in the

future, as will be discussed in Chapter S. In using the xenon-arc lamp

we have found it essential to consider operating procedures to attain a

stable signal. The light source must be run at a minimum of 30 amps and

requires a warm-up time of 20 minutes. Because of the large output in

the ultraviolet, it is necessary, for health reasons, to take some

precautions. Safety eyewear must be worn if, for any reason, the source

is exposed while being used. The system also produces large quantities

of ozone and therefore must be properly vented.

To increase the signal-to-noise ratio of the system, the light

source is modulated by a four-sector chopper which supplies a reference

signal to a lock-in amplifier. This allows us to subtract the dark

signal from the detector.

The Optical System

Beyond the chopper, the light is projected by a lens into the

chamber via a port in the baseplate, where it is laterally displaced by

the two mirrors mentioned previously. The light is then directed

through the reference sample and out of the chamber by an upper window.

A rotating fixture, which is capable of moving more than one sample in

and out of the beam, was added to allow a simple" method for in situ

comparisons. Figure 3.3 depicts the overall arrangement of the scanning

monochromator with respect to the coater. For simplicity, the

36

FRONT VlE.W

Figure 3.3. Appearance of Scanning Monochromator System

37

displacement mirrors at the bottom of the chamber are not shown.

After exiting the coating chamber, the beam is turned by a flat

aluminum mirror and then focused by a lens onto the monochromator

entrance slit, which is 1 cm high and approximately 30 microns wide.

The beam then illuminates the dispersive element of the system, a

concave holographic grating. We use a Joban-Yvon reflective grating,

with 300 lines/mm, designed to disperse light from 400 to 800 nm. The

system was designed such that the entrance slit and detector array lie

upon the Rowland Circle of the grating (Born and Wolfe), which in this

case means the entrance slit should be 21 cm from the grating. The

grating then images the spectrum of the entrance slit onto the CCD

array. When using the negative first order of the holographic grating,

the grating produces a one inch flat field (Lerner et al. 1980). A view

of this part of the system is given in Figure 3.4

The Detector

The scanning monochromator system incorporates a Fairchild Model

122 CCD array as detector with 1728 elements, spaced 26 microns apart.

1728 data points are excessive, and therefore we chose to average this

information in groups of ten pixels, giving us 173 data points through

each scan. (The first and last member of the data set are only averaged

over nine pixels.) The readout electronics for the detector were also

purchased from Fairchild (Model 122DC).

cco ARRAY

TOP VIEW

Figure 3.4. Top View of Scanning Monochromator System w 00

39

Co.mputer Handling o.f Data

The data levels transmitted fro.m the CCO array are sent to. a

dedicated IBM PC, which reco.rds them o.n 5 inch flo.ppy disks and also.

displays them o.n an Amdek video. mo.nito.r fo.r real-time feedback to. the

plant o.perato.r. At the same time, info.rmatio.n fro.m the o.riginal pro.cess

co.ntro.ller, based upo.n the quartz crystal mo.nito.r signal, is sent

thro.ugh an interface mo.dule to. the IBM, where it is also. sto.red and

displayed fo.r the o.perato.r. A flo.wchart sho.wing ho.w the co.mputer

handles the data appears in Figure 3.5. The actual pro.grams fo.r do.ing

this are co.ntained in Appendix A.

The IBM uses a Tecmar analo.g to. digital bo.ard, which accepts

12-bit data fro.m bo.th the quartz crystal and the CCO array. Altho.ugh

the electro.nics are capable o.f running at a rate o.f fo.ur spectra per

seco.nd, we generally o.nly reco.rd a spectrum o.nce every three seco.nds.

The speed o.f the system is currently limited by the rate the co.mputer

system is able to. generate the desired graphics. This ra te o.f o.nce

every three seco.nds is quite adequate fo.r o.ur general purpo.ses. Our

typical depo.sitio.n rates gives us appro.ximately ten to. twenty angstro.ms

between each data acquisitio.n. The po.tential fo.r data rates an o.rder o.f

magnitude faster wo.uld allo.w us to. mo.nito.r ext!'~mely rapid changes

sho.uld the need arise, but wo.uld imply less graphic info.rmatio.n. A

Po.ssible so.lutio.n might be to. limit the graphics to. o.nce every ten runs.

This Wo.uld no.t severely hinder the o.perato.r's view o.f the depo.sitio.n and

Wo.uld allo.w fo.r much faster data acquisitio.n.

STORE

DATA IN

ARRAY

CALCULATE

TRANSMIS-

SION (CCD/-

CCO ARRAY)

TRANS vs WAVE-

>--I~ LENGTH

TRANS va

TIME

SINGLE A, XTAL. TIME

XTAL. TIME

Figure 3.5. Flow Chart of Computer Data Handling Program

40

41

Overall System Performance

Table 3.1 summarizes important characteristics of the

spectrometer part of the system. The minimum transmission which the

system is capable of measuring is defined as a reading of twice the

typical dark signal of the array.

Table 3.1. System Performance

Wavelength Range 440-800 nm

Wavelength Resolution 2nm

Resolving Power 300

Etendu' of System 1 x 10-3 cm2-sr

Minimum Transmission 2 percent

Signal Level of CCD at 100 % Trans 70 % of saturation at 650 nm

42

The system is periodically calibrated in wavelength by inserting a piece

of didymium glass in the light path and then comparing the known

features of its transmission curves with those obtained from our CCD

data. Figure 3.6 shows a Cary 14 transmission curve for the didymium

glass. From this curve we choose seven calibration points, which occtir

at 455, 493, 548, 615, 650, 680, and at 705 nm. Figure 3.7 is a plot of

known wavelength against pixel number it is seen by. This curve

confirms that the system is linear in wavelength. From calibration

curves such as this, or by simple linear interpolation, we are able to

relate pixels to wavelength to an accuracy of approximately two

nanometers.

Applications of the System

Having described the system, we now consider some applications.

In this short discussion we include one relevant application that has not

been carried out in any manner here at the University of Arizona. Of the

applications covered the author has been most closely related to the use

of the system for derivations of optical constants of dielectric layers.

Therefore this application will be mentioned only briefly in this chapter

with a detailed discussion in Chapter 4.

A major current thrust in thin film research is directed towards

materials. A principal interest at the Optical Sciences Center has been

in the visible region of the electromagnetic spectrum.

scanning monochromator has become an invaluable tool.

In this, the

Its strength

stems from its great ability to both take and store data as a permanent

record of any experiment performed in the Balzers 760. As mentioned

~ : I - . ··--

-· .. ·- . . ·· · ·· ...... ··- .

,, ,. .. -· -· --·· .. . .- -- ···· · - ·- - -- · -·-- -- -·-- 1---- -~ ·--· -- - -·· . - - · -· -· - -- - •!: ~ -- ··-· ·- · -- .•. . _..._ _ ll ' l

- ~-- ~- ~ .. Cl.l - - .. 'h

. II ! ~ - !':~ . .:_ -- --- -- -- --- ·~ 'j .

. . - . .

'-- ·--· - - -·- · - -·-- ··-· f--- ,_ ·- ··· - r--- · ...... -· ·-_ _ ·J ·_-· __ · ~- ----· -. .. -. ·.-.- _-__ ·- ·---· . ~-. ·.· (\_._· .·· .-. ·-_ -__ -_ , ____ ~-.. . · ··- 1- -- -- · - · - ~.. : ~- . .:.:_ - - ~- -- ·- t--- .J. - --- ·-·- -- --

'-' -- -- ~-- - r- ~ = ~~ £":, _c 2 --= ~~ = \l~ ~ -= = IJ.j ~ -- - ·- . - -- ··--o.:• --- ·- - - -- - · ~~ -. - - ·· --- --- -- ---- - -·-- ----- ··--- --

f--. - ··- --· - - · -- -·-- --·- - - · · · -- · · · -- --·- --· · ·-· · . -· ..

. 0 . 1 - --- -·- -- ·· .. - --· -- - --- -- ·-· · ~ I - - ·-:- -- f-~J _:~ ~ ~; _::__ .:: ~·- ~- -~ 1-- - -·- . .....: .f- - --- - . ...:.. --· ~ - !'I\ . -\_f K j u . -\ -,_ - ' ' - - -~- _; +~ ~~ -- "-' --,-... , ... ---... ---I .

lP . . \ oj -·- ... .. ... - - . . . o.o ---· - - ·- ··· - - ·-·- ··· - · ----· - - · --·~ ---f--- - · - - r--r-- - - -·- -- -~ · ·--- -·- .. . -· -·· ·-·· .

0 .0

Figure 3.6. Cary 14 Transmission Trace of Didymium Glass

44

160

120

CCD Pixel ~O Number

~O

0 ~oo 500 600 700 aoo

Wavelength (nm)

Figure 3.7. Wavelength Calibration Plot

45

earlier, any deposition of a thin film inherently involves many process

parameters, from the type of starting material to the substrate

temperature and vacuum conditions. Because of the multiplicity of

parameters in any deposition, it is immensely important that as complete

a data base as possible be recorded. This allows for scrupulous

dissection of the data at a later time. The scanning monochromator is

currently able to store approximately 50 minutes of continuously

acquired data on a 5-inch floppy disk for later analysis or comparison.

One of the most frustrating of all the problems inherent in thin

film multilayer coatings is the shift in the spectral performance upon

moving a test sample from the vacuum chamber into its intended

environment. As mentioned in Chapter 1, this shift is a consequence of

the microstructure of the thin film layers. The voids in the columnar

structure of most thin films permit the adsorption water vapor from the

atmosphere. This causes an effective change in the indices of the

layers, which leads to a change in optical thicknesses and thus a change

in the spectral performance of the coating. In many cases, e.g., beam

splitters, this will not severely hinder the use of the coating, though

the addition of the water into the layer may have serious implications

for its lifetime. In the case of films designed for a particularly

narrow wavelength region, the effect can be devastating. An example of

just how drastic this problem can be is given in Figure 3.8. This

example shows a narrow-band Fabry-Perot filter, in which the spacer

layer has adsorbed enough water upon the entrance of air into the vacuum

chamber to shift the narrow peak such that it is no longer in the

original passband. In such a case, the film would have been rendered

w o z <

46

100~-----------------------------------------

SHIFT IN WAVELENGTH WITH PUMPDOWN FOR A NARROW BAND FILTER

f' I I ' I I ' I I , I I , ,

~ 50 , , I , \ I :E

(J) z < a: ....

IN VACUUM I

~ IN AIR : :V J , , I , \ J

, I , I \ I \ I

_/

O -~ \ _/ ~~-'~~~~-T~~T-'--r~----~~------~

440 WAVELENGTH (nm)

880

Figure 3.8. Example of Water Adsorption in Ti02,· Si02 Fabry-Perot

Filter

47

useless. A wide-band transmission monitor reveals such effects much

more quickly. The operator can measure the spectral profile without the

need to first remove the sample. It is also possible to measure the

rate at which the water vapor is adsorbed by the coating, allowing

assessment of attempts to retard water adsorption (Saxe et al., 1985).

Another strength, perhaps its greatest in cases where only final

performance is important, is its ability to provide a figure of merit

during the production of the coating, which contains multispectral

information. In 1980, Bousqet and Pelletier showed, with a computer

simulation, the advantages of monitoring over a wide wavelength region

in the case of a band-pass filter. In their paper it was shown that a

figure of merit of the following form:

fi A2J ITi(A,di) - Ti(A,d)ldA Al

had great advantages over a single wavelength measurement. Acceptable

coatings should be produced in fewer attempts.

The idea of monitoring a coating optically while it is being

sputter-etched-off was suggested by Herrmann and McNeil (1981) and

independently by Botten et al. (1984). Hermann and McNeil's objective

was to investigate layer thickness variations, with Botten et al. it was

the selection of the correct nand k pair out of several possible

solutions for absorbing films. Both of these contributions dealt with a

single wavelength only. We have extended this technique, including its

use over a wide wavelength range, in an exercise organized by the

Optical Society of America. Five groups took part in independently

48

analyzing an unknown multilayer coating. Among other techniques, we

decided to apply wide-band reverse monitoring, that is sputtering off the

coating while optically monitoring. With added information, primarily

consisting of Rutherford Backscattering with some Auger Electron

Spectroscopy, we were able to determine quite correctly the composition

of the coating. An example of a reverse monitoring signal is given in

Figure 3.9. This corresponds to the reverse of what an operator would

have seen while creating the coating, if monitoring at 463nm. Some care

must be taken in analyzing reverse monitoring data. Absolute magnitudes

may at times be somewhat misleading, since the material being sputtered

may tend to have one constituent preferentially removed. This can cause

increased absorption in the film. Our first thought on the information

was to ignore the slight upswing in the data (marked in the figure by an

arrow) since it did not seem reasonable, but after careful consideration

of the results it was determined that the coating did indeed contain one

layer of metal which was only a few angstroms thick. It should be

noted that the scanning monochromator was sensitive enough to show this.

It should also be noted that the data were obtained with the tungsten-

halogen lamp near its lowest radiance level and therefore are examples

of worst-case signals.

Finally, the system can be used to determine the optical

constants of a growing film. Because of the continuous record during

the entire growth of the film it is possible to determine the optical

constants as a function of optical thickness. This is virtually

impossible with conventional techniques which require the film to be

removed from the system. They suffer from effects due to atmospheric

49

Transmission vs. Time

100

50

o

Time (arbitrary units)

Figure 3.9. Example of Reverse Monitoring Trace

50

conditions, but more importantly, the detailed profile of refractive

index through a layer has little effect on its optical characteristics.

Post-deposition measurements are therefore useful only for determining

the total variation of index and inappropriate for examining the profile.

The SMS also permits the measurement of dispersion of the optical

constants and their profile with wavelength. In the following chapter I

will discuss this in more detail.

CHAPTER 4

DERIVATION OF OPTICAL CONSTANTS FROM SMS DATA

In the production of thin film coatings it is important to be

able to characterize each layer deposited. In optical coatings, the

essential information is the index of refraction (n) and extinction

coefficient (k) of the material. Unfortunately the optical constants for

a thin film, nand k, are quite different than the values for the bulk

form of the material. This difference is largely caused by the

structure of the film, i.e., their columnar growth and the

inhomogeneities found within them. Many techniques exist for the

determination of the optical constants of a thin film. Most of these

suffer from the disadvantage of being carried out under atmospheric

conditions, allowing the penetration of water vapour and possible

changes in oxidation states to occur. It is almost impossible to derive

the depth profile of the index of refraction and the extinction

coefficient of the layer. These limitations enormously complicate the

task of understanding the effects of various deposition parameters, to

say nothing of understanding the layer thickness effects.

For these reasons the scanning monochromator system was

developed. The ability of the system to generate wide-band transmission

data of the film at relatively short time intervals during film

deposition makes possible techniques for deriving the depth profiles of a

layer's optical constants. Our method of analysis is based on an

51

52

envelope method devised by Manafacier, Gasiot and Fi11ard in 1976. Their

paper deals with a simple homogeneous layer; the nand k values versus

wavelength were derived from transmission versus wavelength data only,

for a weakly absorbing film. The key assumption was the constancy of d,

the film thickness, at all wave1engths.- In our approach, we vary

wavelength and thickness, making no assumption regarding film

homogeneity.

The Envelope Method

Manafacier, et a1., developed the envelope method for the

limiting case of a weakly absorbing homogeneous layer. A generalization

of this to an inhomogeneous model was presented by Arendt et ale (1984)

but required both reflection and transmission data. In all of these

studies, reflection and transmission are treated as functions of

wavelength at a particular thickness. To obtain the depth profiles of

the optical constants in our study we considered the envelope of our

transmission curves as functions of thickness at a particular

wavelength. We are also able to determine the wavelength dispersion of

the optical constants by compiling the information we have determined at

a chosen thickness.

A 10ssless inhomogeneous layer, where the inhomogeneity can be

considered only a perturbation of an otherwise homogeneous behaviour and

not the dominant factor, can be represented by the following

characteristic matrix (Jacobson, 1975):

53

isin6

where nout is the index at the outer surface(away from the substrate) of

the film, nin is the index at the inner surface of the film, and 15 is the

phase thickness of the film.

If we can assume the absorption in the layer is very small, the

amplitude reflection coefficients at the interface are scarcely affected

if we neglect it. The absorption in the layer is primarily determined by

the phase thickness; consequently, we can include the effects of small

extinction coefficient k by inserting it into 15 so that :

1 rd

where n is the mean index of the film (i.e. n=(Ci) Jo n(z)dz),

k is t he mean extinc tion coe f ficien t. (i.e. k-( ~ ) J : k( z)d z).

and A is the wavelength and d is the thickness of the layer.

If we assume that the incident medium has a refractive index no

and that the substrate is nonabsorbing and of index ns , we find the

overall transmission of the coated substrate is:

where

54

6 = 21fnd 1 A

It should be noted that C2

can be either positive or negative, depending

upon the index of the material one is inves~igating. In the case of a

high index material, discussed here, C2 is negative. The expression of

the envelope of Tmax and Tmin are then given by:

and

Determination of n(d)

We can now use these equations to determine the outermost index

at each instant during deposition, assuming the layer is stable so that

we can consider nin to be constant throughout the deposition process.

n1n:

IoIhere

By setting nin=nout for d=O we find the following expression for

n 2+n 2 N= 0 s

2 Tmax-Tmin

+2nons< T T') max m~n

the expression for nout is given by:

where

Tmax-Tmin Q =

TmaxTmin

55

which is only calculable if we know the innermost index. From these two

expressions we are able to determine the index profile with respect to

thickness, provided that we know the index of refraction of the

substrate as measured independently on an Abbe' Refractometer.

Determination of thickness d

To determine the thickness of the studied film, we make use of

the fact that the extremums in the transmission data occur at

quarterwave optical thickness intervals. (This assumes we are dealing

with weakly absorbing layers). Therefore if m is the order of the

extremum (m=l indicating the first minimum, in the case of a high index

layer) the physical thickness d can be determined by:

d

since the calculation of n is impossible, since d is unknown, we make the assumption that

n - (i) J: n(u)du where t is the instant (time) when the extremum occurs. (This is

effectively assuming that our rate of deposition is constant, which is

reasonable since the system process controller is designed to keep a

56

constant rate.)

Determination of k

To determine the profile of the extinction coefficient we first

note that a mean value of the extinction coefficient can be derived each

time we reach a new quarterwave point:

- A k = -(4~d)loga(d)

where

Since this function a(d) is available at any time during the growth of

the layer, we can calculate its derivative versus time and thus obtain

the extinction coefficient profile:

k(z) = (_ A)( 1 )(da) 4ir arzJ dz

Because of the derivative in the expression this technique is very

sensitive to errors, but it can still be used to obtain an indication of

the absorption of the material.

Smoothing of Transmission of Data

Since our analysis technique is dependent upon an accurate

drawing of the Tmin and Tmax envelope, it is very important to eliminate

any erroneous maximums or minimums due to system noise from our data

base. We need to smooth our raw data before we apply our analysis. In

this smoothing process it is also very important not to distort the

curve in any way. We therefore chose a finite impulse response filter

57

which has a linear phase and a flat low-pass band. In so doing we avoid

any distortion caused by a nonlinear phase and any attenuation due to a

nonflat pass band. The only real artifact of the filter is a constant

delay it introduces; this can be eliminated by an overall shift of the

transmission data. The actual shape of the filter we employ is shown in

figure 4.1. (For further detail on the filters design and performance,

see the text by S. M. Bozic). An example of a signal before and after

filtering is given in Fig 4.2 and 4.3 respectively.

Drawing of Envelope

After the smoothing operation, the extremes of the signal are

easily found. In practice they are located by simply looking for a

change in the sign of the slope of the curve. If the slope changes from

positive to negative, we know that we have just passed through a maximum

in the curve. A negative to positive change indicates a minimum.

Once these extremes are located, some care must be taken in

determining the curves for Tmin and Tmax. These curves are drawn in

segments in the following manner. The firs t group of three points,

corresponding to the first three maxima or the first three minima

depending upon which curve is being drawn, defines a parabola which we

use to describe the segment of the curve joining the first two points.

Next, the first point is discarded, while the fourth point is introduced.

The same process is then used to draw the curve between points two and

three. This process is then repeated until the entire curve has been

produced.

58

0.20

0.15

0.10

WEIGHT 0.05

0.00

-0.05 o 5 1 a 15 20 25 30 35 40 45 50 55

DATA POINT

Figure 4.1. Kalman Filter Used to Smooth Data

59

TRANSMISSION va TIME FOR WAVELENGTH 803 nm 100 .....-----..

60

o~--------------------------------------------TIME

Figure 4.2. Plot of Noisy Signal

60

100 a--.._ FILTERED TRANSMISSION

60

o~---------------------------------TIME

Figure 4.3. Filtered Signal

Computer Simulation of Thin Film Deposition

and Application of Analysis Technique

61

Since several assumptions are inherent in our analysis technique,

we felt we should first test the technique with a computer simulation of

a thin fi,lm with known and regular dispersion of its optical constants

with respect to thickness. We used a fairly standard process of

modeling a single thin film as a film consisting of many even thinner

layers of homogeneous parallel-sided material. In our simulation, we

chose an optical thickness of 1 nm for each sublayer. This was then

processed to determine the transmittance of the layer as each sublayer

was added~ The data were then stored on disk in the same format as the

data generated by the scanning monochromator system. It was then

possible to apply our analysis program to the data directly and, compare

the calculated optical constants information with the known nand k

values.

We did this for two different cases, first the simple case of a

homogeneous layer of index 2.3 and extinction coefficient of 10-3 , and

secondly a layer with dispersion only in its index of refraction, given

as:

n(d) = 2-0.3exp(- zgo)

where d is the thickness measured in nanometers and the constant

extinction coefficient k=10-s• The analysis was then carried out for two

different wavelengths, 400 and 800 nms.

In the case of the homogeneous layer, we derived an index within

0.03 % of the known value. As expected, the calculated extinction

coefficient showed poorer agreement, with a relative error of up to 10 % •

62

For the inhomogeneous layer, we once again found very good agreement in

the index of refraction and less usefu:. though still informative,

extinction coefficient information.

layer are presented in Table 4.1.

The results for the inhomogeneous

Application of Technique to Titania Films

The analysis described in the preceding chapters has been used

to study some of the deposition parameters for titanium films. Several

parameters are particularly important in the deposition of Ti02 on glass

substrates. Of special concern were the temperature of the substrate

and partial pressure of oxygen backfilling the vaccum chamber. The

residual oxygen atmosphere is important because of the tendency of

titanium to lose oxygen during evaporation. Ti203 is considered the

stable form which does not reduce further. Evaporation of Tiz03 demands

additional oxygen to create TiO z•

First the TiOz was deposited in a chamber with an oxygen partial

pressure of 4.4x10-~ mbar and substrate temperature ranging from 204° to

227°C. The physical thickness of the layer deposited under these

conditions was approximately 670 nm. The inner index of the material

was found to be 2.135, while the outer index was 1.794. The actual

profiles of the index and the extinction coefficient for the layer are

given in figure 4.4. The inhomogeneity in the layer can be explained by

a decreasing packing density as one moves out from the substrate, as

well as an increase in the oxidation state nearer the surface. This

difference in oxidation is also demonstrated in Fig. 4.5 in which the

dispersion of the refractive index is versus to the wavelength for the

3.00 0.10

WAVELENGTH 878 NY

2.28 0.0&

N K

1.6J : I I I • I I 10 .00 TIME (THICKNESS)

Figure 4.4. Profile of Refractive Index and Extinction Coefficient for a Stable Titania Layer. (Upper curve represents N, lower

curve K) 0-. W

2.3

2.0

•

REFRACTIVE INDEX

.. ..

•

INNERMOST REFRACTIVE INDEX

• • .. . t t

OUTERMOST REFRACTIVE INDEX

• • • • • • , , '! ' 1.7 .

600 WAVELENGTH (nm)

800

Figure 4.5. Disperion of Innermost and Outermost Refractive Index for a Stable Layer of Titania Film

CT> .I>-

0.10

n WAVELENGTH 678 nm k

2.26 O.Oft

n ,

~ 1.52 I I 0.00

. Figure 4.6. Example of Result Given by Method when Applied to an Unstable Layer

0\ lJ1

66

inner and outer areas of the coating. The profiles of these clearly

show that there is a significant difference between these two portions

of the coating, greater than could be explained simply by a packing

density argument.

A second example concerns a TiOz film which was deposited in a

manner in which one would expect a poorer film to result. In this trial

the partial pressure of oxygen was 1.3xl0-~ mbar and the temperature of

the substrate was held near 260°C. We were aware that the final film

would probably be oxygen deficient and therefore more absorbing. Figure

4.5 indicates that we were correct. For correct application of the

analysis, the film must be stable during the entire process. An

apparent breakdown in this method can therefore be used as an indication

of layer instability. In this case we derive unrealistic values of k,

clear evidence of a lack of stability in the film, that is, a change in

optical constants after deposition. Our analysis assumes the inner index

of the film never changes, thus the program needs to calculate

exaggerated values for the extinction coefficient in order to reconcile

the transmission data with the layer model. It is suspected that this

film instability is caused by the continuing oxidation of the inner

~'~L·iace during the deposition process.

Limitations of the Technique

The most important limitation to this technique is the necessity

for stability of the growing film for the interpretation of

transmittance data in terms of variation of refractive index, n, and

extinction coefficient, k, is to be valid. This limitation is not

67

entirely detrimental. Clear lack of validity of the results is an

indication of layer instability, and a probe of layer instability can be

very useful. However, recognition of instability is all that is possible.

t-iore details of the way in which the properties change are beyond this

technique. For a complete description of the layer, we must add

reflectance measurements which are extremely difficult in a coating

plant. Another problem involved with this technique is the drawing of

the envelopes, discussed in section 4.7. The envelopes dictate the values

of (n,k) that will be determined. It is important, therefore, that the

turning points in the transmission curve be properly located. To ensure

this, it is necessary to have a reasonable contrast in index between the

substrate and the film.

CHAPTER 5

SUMMARY

With the description of the scanning monochromator system

complete I conclude with a brief summary of what we have accomplished

in our lab at the Optical Sciences Center at the University of Arizona

and a description of a system which we are currently building to extend

our research into the ultraviolet region of the spectrum. There will

also be a brief discussion of future experiments which may be done with

the system by those who carryon this work.

A Scanning Monochromator System has been built to further the

ability of monitoring thin film deposition and to increase our

capability of analysing the microstructure of the film deposited in the

chamber. To this date the system· has been used in several different

experiments to analyse the properties of a film, but has not yet been

fully utilized in its monitoring capabilities within our laboratory.

This is primarily attributable to the emphasis of our reseach group,

which is, to a much larger extent, to investigate thin film material

properties rather than the production of specialized optical filters.

This system offers several adyantages over con